Applied AI Architect · HealthTech Strategist · Independent Researcher

AI systems built to survive contact with reality.

DeNovo Innovations Group architects, builds, and evaluates applied AI and healthcare technology systems where evidence, provenance, human judgment, and reliability matter.

Explore the workView research & evidence

Continuous Behavioral Access Management

CBAM — Continuous Behavioral Access Management: concentric trust spheres from the untrusted web through tools and MCP servers, the Internet of Agents with private agent clouds, agent networks and memory networks, to a governance and proof layer.
Proposed concentric trust spheres and system architectureResearch vision: Continuous Behavioral Access Management and the Internet of Agents.

Proof

Three things to look at first

Healthcare policy intelligenceWhich policy revision applied, and when — change-aware retrieval over manuals with hundreds of revisions.Read more →
AllMyMascotsBuilt end to end by a human-AI hybrid system: a production generative-AI platform with multi-provider inference, failover, preference learning and evidence-gated release.Read more →
The DeNovo toolkit
Diagram of the DeNovo human-AI engineering system: human direction, role-separated agents, repository context, verifiers, hooks and retained evidence; solid borders mark implemented parts and dashed borders planned ones.
The DeNovo toolkit. Solid border = implemented, dashed = planned.See it on Evidence & Provenance →
Public comments to NIST and FDAEvaluator integrity, dated evidence and action-level assurance — DOI-archived contributions on AI evaluation.Read more →

How I think

From electrons to evidence.

Medicine first, then every layer of infrastructure it runs on, then the data, the models and the AI systems that reason over them. Every layer below is work performed, not a reading list; failures live at the seams between layers, and that is where I look first.

  1. 01Medicine & clinical researchBiomedical engineering; cardiothoracic surgery outcomes research; three years of clinical training
  2. 02Electrical & physical systemsPower distribution and controls during the decommissioning of an automated distribution center; sensing hardware built by hand
  3. 03Networks & connected environmentsDistrict-scale wireless infrastructure; unified communications for manufacturers; IoT and control systems for intelligent buildings
  4. 04Sensing & edgeCamera-based vital-sign rigs on edge compute, integrated with robotics and deployed in the field
  5. 05Data & cloudSecure Microsoft 365 and Azure environments for clinical research; least-privilege identity; hashed, timestamped evidence of system state
  6. 06Machine learningReal-time sleep-disturbance detection from wearable signals; rPPG heart-rate and breathing-rate extraction from webcam video
  7. 07AI systems & evidenceRevision-aware policy retrieval; a production generative-AI platform; public comments to NIST and FDA on evaluation evidence and provenance

The four habits that travel between those layers →

How AI systems are built here

One architecture, eleven questions

  1. 01identity / authorization→
  2. 02epistemic scope→
  3. 03authority resolution→
  4. 04temporal resolution→
  5. 05context compilation→
  6. 06reasoning / deterministic execution→
  7. 07verification→
  8. 08provenance→
  9. 09durable knowledge→
  10. 10memory audit→
  11. 11counterfactual audit

Every AI system answers these in order; where it cannot, it says so. How claims on this site are labelled

Selected work

Built, deployed, written down

Contactless vital signs from videoMedTech · computer vision and signal processing; rigs built, deployed, acquired.Read more →Remote patient monitoring programMedTech · wearables, reimbursement model, pre-submission planning toward a Class II pathway.Read more →Research programTwenty-five topics in four programs, each labelled by maturity.Read more →

Work With Me

Selective collaboration where applied AI, healthcare technology, evidence, and real-world implementation intersect.

Healthcare is the focus, not the boundary. Same problems, every discipline; let's solve them together.

Engagements run through DeNovo Innovations Group. Write with what you are working on and the decision it supports.

Engagement lanes and contactgennaro.maida@denovoinnovationsgroup.com